AWRA banner

Advancing Water Resources Research and Management

AWRA SYMPOSIUM ON GIS AND WATER RESOURCES
Sept 22-26, 1996
Ft. Lauderdale, FL

USE OF ARC/INFO, EPA-SWMM, AND UNIX TEXT PROCESSING TOOLS TO DETERMINE FLOOD EXTENT

T.B. Cera¹, T.K. Tremwel², and R.W. Burleson³

Abstract
Introduction
Objectives
Methods
GIS Data
EPA-SWMM Model
Discussion and Conclusion
References
Acknowldegements

Abstract

This paper presents work performed for the Deland Ridge Watershed Management Plan for Volusia County, Florida. The 60 square mile study area consists of 280 subcatchments. None of the subcatchments have any apparent surface hydraulic connection to neighboring subcatchments. Input data for the EPA-SWMM model was developed from the soils, land use, and elevation contour coverages. The EPA-SWMM model was run for both the present and planned build-out land use condition, using a 100-year, 24-hour storm event. The maximum flood elevation within each subcatchment was extracted from the model output using UNIX text processing tools. An ARC/INFO AML script was created to take the maximum flood elevation results and assign all grid cells in the elevation grid below that elevation a special code. That code is used to display the flooded areas in the desired color, allowing comparisons to mapped roads and structures. The flood maps show the influence of increased urbanization on runoff volumes and identified subcatchments with particular drainage problems.

Introduction

The work outlined in this paper was performed as part of the Deland Ridge Watershed Management Plan for Volusia County, Florida. The project objective was to identify stormwater quantity and quality issues in the watershed and to provide a valuable planning tool for future improvements to the stormwater management system. ARC/INFO Geographic Information System (GIS) with the GRID module was used as a pre and post processor to the Environmental Protection Agency - Storm Water Management Model (EPA-SWMM). Text and data processing tools that are a standard part of the UNIX operating system were used to manipulate the model output data into a form suitable for ARC/INFO. The intent of the project was to effectively model a very large watershed with intensive data requirements in a short span of time. Though ARC/INFO was used as a pre and post processor to EPA-SWMM, it was never set-up as a connected, interactive tool.

Volusia county is on the east coast of central Florida and is bordered on the west by the St. Johns River. The Deland Ridge Watershed, almost on the western edge of the county, has a total area of approximately 61.5 square miles (159 square kilometers) consisting of 280 subcatchments. Municipalities in the watershed include: DeLand, Orange City, and DeBary. Elevations in the watershed range from 115 feet (35 meters) to 10 feet (3 meters) NGVD. The geology is karstic with much of the area characterized by high local relief, sinkhole lakes, ponds, and dry depressions. There are no major surface water drainage systems in the watershed. Nearly all of the subcatchments drain stormwater by recharge to the surficial groundwater supply. Estimates of the recharge rate for this watershed range from 10 to 18 inches per year (ATM, 1995). This recharge to the groundwater supply was one of the motivations for developing a watershed management plan. As an analogy, each subcatchment is a separate cup when filled with rain slowly drains through the bottom of the cup into the ground. The separateness of the subcatchments is the primary reason why this method of identifying flood prone areas worked. This "cup" like characteristic both simplified the surface water modeling efforts and allowed a single flood stage elevation value to identify the flood stage for the entire subcatchment. Subcatchment delineations were digitized from USGS quadrangle maps. Figure 1 shows the Deland Ridge watershed with subcatchments defined and area municipalities.

Figure 1. Site location, watershed, and

subcatchment boundaries.

The GIS and data processing work for this paper was done on a SUN SPARCstation LX running the Solaris 2.1 operating system with ARC/INFO 7.0.2 and the ARC/INFO GRID module. The GRID module implements a cell based GIS system along with many manipulation functions and tools. A cell based GIS is a grid of cells where each cell contains data and the grid is spatially referenced within the GIS. This version of GRID included the TOPOGRID tool, which proved to be invaluable to the development of the elevation surface. ARC/INFO has a complete internal programming script called Arc Macro Language (AML). An AML script can control any ARC/INFO process saving the user from repetitive work and error. There is a bug in AML scripts; you cannot use a tab to format lines. One tab anywhere in the file will make the AML fail.

Objectives

The objectives of the project were as follows:

Use ARC/INFO as a pre-processor for data input needed by EPA-SWMM.
- Weighted-area per subcatchment soil and land use data.
- Stage-area data.
Use ARC/INFO as a visualization tool of EPA-SWMM output.
Use UNIX text processing tools for data reduction and management.
Compare flooding under present and future build-out land uses.

The objective of this paper is to describe a process for integrating ARC/INFO with EPA-SWMM and UNIX tools to create a system to rapidly model large projects. Primary focus for this paper is placed on the "nuts and bolts" of the project. The intent is to illustrate the process of getting answers from GIS appropriate for water resources modeling and development of visualization of model output.

Methods

Two UNIX text processing tools were used in this study and a short description of each is shown in Table 1. The text processing tools used were controlled using a UNIX Bourne shell script. The Bourne shell's powerful scripting language allows execution and control of all programs available on UNIX platforms. Grep and awk programs are two of the powerful tools that come standard on UNIX platforms. A invaluable part of the application of these programs to the modeling effort was the book by Peek, et al. (1994).

Table 1. Grep and awk text processing programs.
grep	`grep DSEU001 datafile.out`	Prints only the lines in datafile.out that have DSEU001 anywhere on the line.	general regular expression parser. Also claimed to be named from a common series of commands in the UNIX ed line editor (/g/r/e/p).
awk	`awk '/^9/{print $2, $1}' datafile.out`	Prints the second, then first fields of each line in datafile.out that begins with a 9.	awk is named after its developers, Aho, Weinberger, and Kernighan. It is a powerful text processing language with a very terse syntax.

Awk has several unique and powerful features. Unlike most editors awk operates line by line. This allows awk to work on any size file since only one line of data is processed at a time. Awk programs can contain three sections; BEGIN, body, and END. All statements in BEGIN are performed before any data is processed. Body statements are executed for each line in the data file and END statements are executed after the data lines have been processed.

UNIX shell scripts can be very confusing, especially scripts that include awk commands. There are similar programming syntax in UNIX shell scripts and awk that do very different things according to the immediate context. The inclusion of two (or more) distinct programming languages in the same file can be disheartening to the novice user, but there is an incredible power in the combination of UNIX shell scripts and text processing programs with other utilities and models.

GIS Data

Soils, land use, and road coverages are from the Volusia County GIS Department. Soils data are based on the SCS Soil Survey of Volusia County. Almost all of the land area consisted of soils in the 'A' SCS hydrologic group, implying low run-off potential. Present land use data are from a vegetation coverage supplied by Volusia County, 1993 REDI maps, and from field visits. The future land use conditions are from the Volusia County Comprehensive Plan for the year 2010. Since the future land use information was developed from a zoning perspective, it was modified using ARC/INFO to account for the actual potential build-out of the land. Modifications included taking water and wetland uses from the present land use coverage and adding them to the future land use.

The elevation contours were digitized from USGS 1:24,000 quadrangles and converted into a polygon coverage with an elevation attribute. This is not typical and was done in order to establish stage-area relationships for input into the storm water models. Elevation coverages usually have arc topology with the elevation assigned as an attribute of the arc. In fact, arc topology is what is required by the TOPOGRID tool to develop an elevation surface. The attributes in the polygon elevation coverage were assigned to the correct arc by a complex database manipulation. In hindsite there were other methods that could have been used to develop the stage-area values from the elevation surface, thereby eliminating the need to have the elevation contours be digitized as a polygon coverage. A comparison of the initial polygon topology and the converted arc topology is shown in Figure 2. TOPOGRIDTOOL was used to form the elevation surface from the arc elevation coverage. The elevation surface was created with 50 foot (15.24 meter) by 50 foot cells, with each cell containing an elevation value. This cell size is an appropriate level of detail for the project.

Figure 2. Uses and comparison between

polygon and arc topology and attributes.

EPA-SWMM Model

Data required for the EPA-SWMM model was taken from several coverages. The soil and land use characteristics were developed as area weighted averages per subcatchment using the ARC/INFO STATISTICS command. The stage-area values were developed from the topographic coverage that contained polygons with elevation attributes. The surface used for the base map was created with the ARC/INFO TOPOGRID tool.

A conceptual diagram of the entire process is presented as Figure 3. This diagram describes in detail the processes used and the interaction between the different components.

Figure 3. Conceptual data flow diagram.

All 280 subcatchments could not be processed in one input data set so subcatchments where split into three groups. A shell script was used to take data from six EPA-SWMM output files (three for present land use and three for build-out land use) and convert it into a form that could be brought into ARC/INFO. In order to make the script (Figure 4) easier to read the entire awk program is set to a shell variable. A shell "for" command loops through all of the subcatchments in turn. Each time the shell script loops it uses grep to search for all lines in the data file that contain the subcatchment name. These lines are sent to the awk command which finds the local maximum and prints the basin name and local maximum to the screen or a file.

Using the data file containing maximum flood stage per subcatchment, an AML script (Figure 5) imported the flood elevation values and compared them with the elevation surface grid to create the flood maps. The "con" function used in the AML manipulates the cell values based upon a logic test. This logic test can use other grids to change values in the cells of the output grid. The AML takes the flood elevation for each subcatchment and sets all cell values below the flood elevation to 200. The number 200 is arbitrary and chosen because it is unique within the elevation surface grid and could be used later to highlight flooded areas.
Figure 4. Shell script using grep and awk to process
EPA-SWMM output files to find maximum flood depth.

#!/bin/sh ## The awkscr variable is set to the entire awk script. This allows ## the awk script to be included in the awk command later on in the ## shell script. The awk script can be included directly after the ## awk command. The method shown is used ONLY to make the overall ## UNIX script more readable. The awk script finds the maximum depth ## in the EPA-SWMM output file. This script finds the first local ## maximum, and prints the basin name and the maximum value on stdout. awkscr=' { if ( old < new || ( old == 0 && new == 0 ) ) { old = new new = $2 } else { print $1, old exit } }' ## The for loop goes through the basin names one by one. Only a part ## of one for loop is shown here for illustration. The grep command ## finds every line the basin name is mentioned and pipes the list ## to the awk command with the script from above. The result is a ## single line printed out showing the basin name and the maximum ## value. for basin in DSEU001 DSEU002 DSEU003 DSEU004 DSEU005 DSEU006 do grep "$basin" depthds100.out | awk "$awkscr" done

Figure 5. AML script to bring in data file of maximum flood depths and create grid with all cells in a subcatchment below the maximum flood elevation set to 200.

/* Open flood elevation data file &s fileunit := [open future.dat openstat -read] /* Loop through all of the basins. &do basin = 1 &to 280 /* Read flood elevation for the basin &s basin%basin% = [read %fileunit% readstat] &type Basin %basin% has a flood elevation of [value basin%basin%] &s ck [exists outgrid1 -GRID] &if %ck% &then kill outgrid1 /* con is a logic operator for grids. Usage is 'con(test, /* if test is true, if test is false)' Many of the grid /* operators work on each cell in a grid. In the following /* statement each cell in outgrid is set to either 200 or /* the value of the corresponding cell in fsurfacegrid, /* depending on the results of the logic tests in the con /* statements. If the cell in basin_grid equals the current /* basin, then if the corresponding cell in fsurfacegrid is /* less than the maximum flood stage, set fsurfacegrid to /* 200, otherwise leave the cell in fsurfacegrid alone. outgrid1 = con(basin_grid == %basin%, ~ con(fsurfacegrid < [value basin%basin%], ~ 200,fsurfacegrid),fsurfacegrid) &s ck [exists fsurfacegrid -GRID] &if %ck% &then kill fsurfacegrid rename outgrid1 fsurfacegrid &end /* Close the input data file. &s closestat := [close -all]

Figure 4. Shell script using grep and awk to process
EPA-SWMM output files to find maximum flood depth.
#!/bin/sh ## The awkscr variable is set to the entire awk script. This allows ## the awk script to be included in the awk command later on in the ## shell script. The awk script can be included directly after the ## awk command. The method shown is used ONLY to make the overall ## UNIX script more readable. The awk script finds the maximum depth ## in the EPA-SWMM output file. This script finds the first local ## maximum, and prints the basin name and the maximum value on stdout. awkscr=' { if ( old < new \|\| ( old == 0 && new == 0 ) ) { old = new new = $2 } else { print $1, old exit } }' ## The for loop goes through the basin names one by one. Only a part ## of one for loop is shown here for illustration. The grep command ## finds every line the basin name is mentioned and pipes the list ## to the awk command with the script from above. The result is a ## single line printed out showing the basin name and the maximum ## value. for basin in DSEU001 DSEU002 DSEU003 DSEU004 DSEU005 DSEU006 do grep "$basin" depthds100.out \| awk "$awkscr" done

Figure 5. AML script to bring in data file of maximum flood depths and create grid with all cells in a subcatchment below the maximum flood elevation set to 200.
/* Open flood elevation data file &s fileunit := [open future.dat openstat -read] /* Loop through all of the basins. &do basin = 1 &to 280 /* Read flood elevation for the basin &s basin%basin% = [read %fileunit% readstat] &type Basin %basin% has a flood elevation of [value basin%basin%] &s ck [exists outgrid1 -GRID] &if %ck% &then kill outgrid1 /* con is a logic operator for grids. Usage is 'con(test, /* if test is true, if test is false)' Many of the grid /* operators work on each cell in a grid. In the following /* statement each cell in outgrid is set to either 200 or /* the value of the corresponding cell in fsurfacegrid, /* depending on the results of the logic tests in the con /* statements. If the cell in basin_grid equals the current /* basin, then if the corresponding cell in fsurfacegrid is /* less than the maximum flood stage, set fsurfacegrid to /* 200, otherwise leave the cell in fsurfacegrid alone. outgrid1 = con(basin_grid == %basin%, ~ con(fsurfacegrid < [value basin%basin%], ~ 200,fsurfacegrid),fsurfacegrid) &s ck [exists fsurfacegrid -GRID] &if %ck% &then kill fsurfacegrid rename outgrid1 fsurfacegrid &end /* Close the input data file. &s closestat := [close -all]

Discussion and Conclusion

The flood map for the entire watershed with the present land use condition is shown in Figure 6. A comparison of the flood maps simulated with present land use and build-out land use conditions is shown in Figure 7. Figure 7 illustrates a few things about the process that was used for visualizing the results of the modeling effort. These maps are at a detailed scale to show the 50 foot (15.24 meter) by 50 foot cells of the elevation grid indicated by the blocked edge of the flooded areas. The flood map could have been converted to a polygon coverage and the flooded area boundary smoothed. There is little advantage to smoothing the boundary aside from aesthetic reasons. This is in comparison to the road and watershed boundary coverages which are arc and polygon coverages respectively and do not show the block effect that exists in grid based GISs. In comparing the present and future build-out flood maps it is apparent that the greatest increase in flood risk occurs where undeveloped land (near the western edge of the maps) is converted to urban. The existing urban land uses do not show much increase in flood potential because they are near the build-out capacity.

The following conclusions can be reached from this work:

ARC/INFO can be an effective tool for taking data from GIS coverages and manipulating it into an appropriate format for a hydrologic or hydraulic model.
Visualization of spatial data in the water resources fields can be done effectively with GIS.
ARC/INFO is suited for large modeling projects with intensive data requirements.
ARC/INFO can be used to compare flooding under present and future build-out land uses.
UNIX text processing tools are effective for data reduction and management.

Figure 6. Flood map from simulation of present land use conditions.

Figure 7. Flood map comparison between present and future build-out land use.

Flooded areas with build-out land uses.

Flooded areas with present land uses.

References

Applied Technology and Management. 1995. DeLand Ridge Watershed Management Plan - Final Engineering Report. Prepared for County of Volusia, Stormwater Utility Management.

Environmental Systems Research Institute. 1994. ARC/INFO 7.0.2 manuals, Redlands, CA.

U.S. Environmental Protection Agency. 1988. Stormwater Management Model, Version 4, User's Manual.

Peek, J., O'Reilly, T., Loukides, M., and other contributors, 1994. UNIX Power Tools. O'Reilly & Associates/Random House, Sebastopol, CA, 1129 pp.

Acknowledgements

The authors would like to acknowledge the assistance of the Volusia County Stormwater Utility Management who funded the watershed management planning effort and the Volusia County GIS department who provided the base ARC/INFO coverages.

Authors

1. T.B. Cera
Associate Engineer/GIS Analyst
KBN Engineering and Applied Sciences/Golder Associates
6241 NW 23rd Street - Suite 500
Phone 352-336-5600
Fax 352-336-6603
Gainesville, FL 32653-1500
Email: tcera@kbn.golder.com

2. T.K. Tremwel
Senior Engineer
Applied Technology and Management
2770 NW 43rd Street - Suite B
Gainesville, FL 32606-7419
Phone 352-375-8700
Fax 352-375-0995

3. R.W. Burleson
Senior Engineer and Associate
Applied Technology and Management
2770 NW 43rd Street - Suite B
Gainesville, FL 32606-7419
Phone 352-375-8700
Fax 352-375-0995

GIS Symposium | AWRA

home page
Maintainer: AWRA Webserver Team Last modified: 24Nov99 gaw
Copyright(c) 1996, American Water Resources Association

AWRA SYMPOSIUM ON GIS AND WATER RESOURCES Sept 22-26, 1996 Ft. Lauderdale, FL